[U-Boot] Resets on mx6qsabrelite

[U-Boot] Resets on mx6qsabrelite

[Fabio Estevam]

Hi,

I am getting reset when using latest U-boot mainline with
mx6qsabrelite and loading a kernel.


U-Boot 2012.07-00580-g49b3962 (Sep 19 2012 - 09:56:23)

CPU:   Freescale i.MX6Q rev1.0 at 792 MHz
Reset cause: POR
Board: MX6Q-Sabre Lite
DRAM:  1 GiB
WARNING: Caches not enabled
MMC:   FSL_SDHC: 0, FSL_SDHC: 1
In:    serial
Out:   serial
Err:   serial
Net:   FEC
Hit any key to stop autoboot:  0
Using FEC device
TFTP from server 192.168.1.103; our IP address is 192.168.1.108
Filename 'uImage'.
Load address: 0x12000000
Loading: #################################################################
         #################################################################
         #################################################################
         #########################
done
Bytes transferred = 3220080 (312270 hex)
Using FEC device
TFTP from server 192.168.1.103; our IP address is 192.168.1.108
Filename 'imx6q-sabrelite.dtb'.
Load address: 0x11ff0000
Loading: #
done
Bytes transferred = 13715 (3593 hex)
## Booting kernel from Legacy Image at 12000000 ...
   Image Name:   Linux-3.5.3
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    3220016 Bytes = 3.1 MiB
   Load Address: 10008000
   Entry Point:  10008000
   Verifying Checksum ... OK
## Flattened Device Tree blob at 11ff0000
   Booting using the fdt blob at 0x11ff0000
   Loading Kernel Image ... OK
OK
   Using Device Tree in place at 11ff0000, end 11ff6592
data abort
pc : [<4ff81fd8>]          lr : [<4ff75950>]
sp : 4fd6fb90  ip : 11ff028c     fp : 00000000
r10: 4ffae980  r9 : 00000000     r8 : 4fd6ff60
r7 : 11ff0000  r6 : 00000000     r5 : 00000000  r4 : 4ffae980
r3 : 00000000  r2 : 4ffa2529     r1 : 4fd6fb88  r0 : 11ff0000
Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
Resetting CPU ...

resetting ...


U-Boot 2012.07-00580-g49b3962 (Sep 19 2012 - 09:56:23)

Does anyone experience the same issue?

Any ideas before I start bisecting this?

Thanks,

Fabio Estevam

[U-Boot] Resets on mx6qsabrelite

[Robert Nelson]

On Wed, Sep 19, 2012 at 7:59 AM, Fabio Estevam <festevam@gmail.com> wrote:
> Hi,
>
> I am getting reset when using latest U-boot mainline with
> mx6qsabrelite and loading a kernel.
>
>
> U-Boot 2012.07-00580-g49b3962 (Sep 19 2012 - 09:56:23)
>
> CPU:   Freescale i.MX6Q rev1.0 at 792 MHz
> Reset cause: POR
> Board: MX6Q-Sabre Lite
> DRAM:  1 GiB
> WARNING: Caches not enabled
> MMC:   FSL_SDHC: 0, FSL_SDHC: 1
> In:    serial
> Out:   serial
> Err:   serial
> Net:   FEC
> Hit any key to stop autoboot:  0
> Using FEC device
> TFTP from server 192.168.1.103; our IP address is 192.168.1.108
> Filename 'uImage'.
> Load address: 0x12000000
> Loading: #################################################################
>          #################################################################
>          #################################################################
>          #########################
> done
> Bytes transferred = 3220080 (312270 hex)
> Using FEC device
> TFTP from server 192.168.1.103; our IP address is 192.168.1.108
> Filename 'imx6q-sabrelite.dtb'.
> Load address: 0x11ff0000
> Loading: #
> done
> Bytes transferred = 13715 (3593 hex)
> ## Booting kernel from Legacy Image at 12000000 ...
>    Image Name:   Linux-3.5.3
>    Image Type:   ARM Linux Kernel Image (uncompressed)
>    Data Size:    3220016 Bytes = 3.1 MiB
>    Load Address: 10008000
>    Entry Point:  10008000
>    Verifying Checksum ... OK
> ## Flattened Device Tree blob at 11ff0000
>    Booting using the fdt blob at 0x11ff0000
>    Loading Kernel Image ... OK
> OK
>    Using Device Tree in place at 11ff0000, end 11ff6592
> data abort
> pc : [<4ff81fd8>]          lr : [<4ff75950>]
> sp : 4fd6fb90  ip : 11ff028c     fp : 00000000
> r10: 4ffae980  r9 : 00000000     r8 : 4fd6ff60
> r7 : 11ff0000  r6 : 00000000     r5 : 00000000  r4 : 4ffae980
> r3 : 00000000  r2 : 4ffa2529     r1 : 4fd6fb88  r0 : 11ff0000
> Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
> Resetting CPU ...
>
> resetting ...

It might be masking the problem, but I've been reverting:

http://git.denx.de/?p=u-boot.git;a=commit;h=c50afc1dab78a1a72bc37b60d97864e56b52cc5c

(I've also been reverting that for mx51evk/mx53loco when using device
tree's too...)

Regards,

-- 
Robert Nelson
http://www.rcn-ee.com/

[U-Boot] Resets on mx6qsabrelite

[Fabio Estevam]

Hi Robert,

On Wed, Sep 19, 2012 at 10:31 AM, Robert Nelson <robertcnelson@gmail.com> wrote:

> It might be masking the problem, but I've been reverting:
>
> http://git.denx.de/?p=u-boot.git;a=commit;h=c50afc1dab78a1a72bc37b60d97864e56b52cc5c
>
> (I've also been reverting that for mx51evk/mx53loco when using device
> tree's too...)

Yes, thanks. I have just confirmed it.

Albert,

Your commit c50afc1dab78a (Revert "arm: armv7: add compile option
-mno-unaligned-access if available") causes issues (seen on i.MX at
least) to boot a kernel:

## Flattened Device Tree blob at 11ff0000
   Booting using the fdt blob at 0x11ff0000
   Loading Kernel Image ... OK
OK
   Using Device Tree in place at 11ff0000, end 11ff6592
data abort
pc : [<4ff81fd8>]          lr : [<4ff75950>]
sp : 4fd6fb90  ip : 11ff028c     fp : 00000000
r10: 4ffae980  r9 : 00000000     r8 : 4fd6ff60
r7 : 11ff0000  r6 : 00000000     r5 : 00000000  r4 : 4ffae980
r3 : 00000000  r2 : 4ffa2529     r1 : 4fd6fb88  r0 : 11ff0000
Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
Resetting CPU ...

resetting ...

Any suggestion?

Regards,

Fabio Estevam

[U-Boot] Resets on mx6qsabrelite

[Tom Rini]

On Wed, Sep 19, 2012 at 11:18:14AM -0300, Fabio Estevam wrote:
> Hi Robert,
> 
> On Wed, Sep 19, 2012 at 10:31 AM, Robert Nelson <robertcnelson@gmail.com> wrote:
> 
> > It might be masking the problem, but I've been reverting:
> >
> > http://git.denx.de/?p=u-boot.git;a=commit;h=c50afc1dab78a1a72bc37b60d97864e56b52cc5c
> >
> > (I've also been reverting that for mx51evk/mx53loco when using device
> > tree's too...)
> 
> Yes, thanks. I have just confirmed it.
> 
> Albert,
> 
> Your commit c50afc1dab78a (Revert "arm: armv7: add compile option
> -mno-unaligned-access if available") causes issues (seen on i.MX at
> least) to boot a kernel:
> 
> ## Flattened Device Tree blob at 11ff0000
>    Booting using the fdt blob at 0x11ff0000
>    Loading Kernel Image ... OK
> OK
>    Using Device Tree in place at 11ff0000, end 11ff6592
> data abort
> pc : [<4ff81fd8>]          lr : [<4ff75950>]
> sp : 4fd6fb90  ip : 11ff028c     fp : 00000000
> r10: 4ffae980  r9 : 00000000     r8 : 4fd6ff60
> r7 : 11ff0000  r6 : 00000000     r5 : 00000000  r4 : 4ffae980
> r3 : 00000000  r2 : 4ffa2529     r1 : 4fd6fb88  r0 : 11ff0000
> Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
> Resetting CPU ...
> 
> resetting ...
> 
> Any suggestion?

Albert intended that as a band-aid for v2012.07 and for the real issues
to be fixed for v2012.10 (and the previous threads about the options
should explain what Albert expects wrt patches).  Thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20120919/68e7a2e2/attachment.pgp>

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Tom Rini,

> On Wed, Sep 19, 2012 at 11:18:14AM -0300, Fabio Estevam wrote:
> > Hi Robert,
> > 
> > On Wed, Sep 19, 2012 at 10:31 AM, Robert Nelson <robertcnelson@gmail.com> 
wrote:
> > > It might be masking the problem, but I've been reverting:
> > > 
> > > http://git.denx.de/?p=u-boot.git;a=commit;h=c50afc1dab78a1a72bc37b60d97
> > > 864e56b52cc5c
> > > 
> > > (I've also been reverting that for mx51evk/mx53loco when using device
> > > tree's too...)
> > 
> > Yes, thanks. I have just confirmed it.
> > 
> > Albert,
> > 
> > Your commit c50afc1dab78a (Revert "arm: armv7: add compile option
> > -mno-unaligned-access if available") causes issues (seen on i.MX at
> > least) to boot a kernel:
> > 
> > ## Flattened Device Tree blob at 11ff0000
> > 
> >    Booting using the fdt blob at 0x11ff0000
> >    Loading Kernel Image ... OK
> > 
> > OK
> > 
> >    Using Device Tree in place at 11ff0000, end 11ff6592
> > 
> > data abort
> > pc : [<4ff81fd8>]          lr : [<4ff75950>]
> > sp : 4fd6fb90  ip : 11ff028c     fp : 00000000
> > r10: 4ffae980  r9 : 00000000     r8 : 4fd6ff60
> > r7 : 11ff0000  r6 : 00000000     r5 : 00000000  r4 : 4ffae980
> > r3 : 00000000  r2 : 4ffa2529     r1 : 4fd6fb88  r0 : 11ff0000
> > Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
> > Resetting CPU ...
> > 
> > resetting ...
> > 
> > Any suggestion?
> 
> Albert intended that as a band-aid for v2012.07 and for the real issues
> to be fixed for v2012.10 (and the previous threads about the options
> should explain what Albert expects wrt patches).  Thanks!

Link would help, thanks ;-)

Best regards,
Marek Vasut

[U-Boot] Resets on mx6qsabrelite

[Fabio Estevam]

On Wed, Sep 19, 2012 at 4:29 PM, Marek Vasut <marex@denx.de> wrote:

> Link would help, thanks ;-)

Yes, please. I was not involved on past discussions on this topic and
I would really like to understand what is the recommended way to
fixing this problem.

Regards,

Fabio Estevam

[U-Boot] Resets on mx6qsabrelite

[Tom Rini]

On Wed, Sep 19, 2012 at 05:13:20PM -0300, Fabio Estevam wrote:
> On Wed, Sep 19, 2012 at 4:29 PM, Marek Vasut <marex@denx.de> wrote:
> 
> > Link would help, thanks ;-)
> 
> Yes, please. I was not involved on past discussions on this topic and
> I would really like to understand what is the recommended way to
> fixing this problem.

I lmgtfy.com'd Marek, but for everyone,
http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the start
of the thread.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20120919/89801e0f/attachment.pgp>

[U-Boot] Resets on mx6qsabrelite

[Fabio Estevam]

On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini <trini@ti.com> wrote:

> I lmgtfy.com'd Marek, but for everyone,
> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the start
> of the thread.

I am still getting the same issue on 2012.10-rc1.

Is anyone else having issues to load a kernel with 2012.10-rc1 on sabrelite?

Thanks,

Fabio Estevam

[U-Boot] Resets on mx6qsabrelite

[Eric Nelson]

On 09/24/2012 05:03 PM, Fabio Estevam wrote:
> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
>
>> I lmgtfy.com'd Marek, but for everyone,
>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the start
>> of the thread.
>
> I am still getting the same issue on 2012.10-rc1.
>
> Is anyone else having issues to load a kernel with 2012.10-rc1 on sabrelite?
>

Hi Fabio,

I'm having some trouble running with cache enabled (on Dirk's branch),
but I wasn't having issues without it

I haven't pulled 2012.10-rc1 and won't be able to test until
late in the week (traveling).

Regards,


Eric

[U-Boot] Resets on mx6qsabrelite

[Tom Rini]

On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini <trini@ti.com> wrote:
> 
> > I lmgtfy.com'd Marek, but for everyone,
> > http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the start
> > of the thread.
> 
> I am still getting the same issue on 2012.10-rc1.
> 
> Is anyone else having issues to load a kernel with 2012.10-rc1 on sabrelite?

I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
Albert, what do you want to do about this?

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20120924/98dd8e60/attachment.pgp>

[U-Boot] Resets on mx6qsabrelite

[Eric Nelson]

On 09/24/2012 05:22 PM, Tom Rini wrote:
> On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
>> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
>>
>>> I lmgtfy.com'd Marek, but for everyone,
>>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the start
>>> of the thread.
>>
>> I am still getting the same issue on 2012.10-rc1.
>>
>> Is anyone else having issues to load a kernel with 2012.10-rc1 on sabrelite?
>
> I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
> on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
> Albert, what do you want to do about this?
>

That adds up, since I've been building with an old one (4.4) and
haven't been seeing resets.

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Eric Nelson,

> On 09/24/2012 05:03 PM, Fabio Estevam wrote:
> > On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
> >> I lmgtfy.com'd Marek, but for everyone,
> >> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the start
> >> of the thread.
> > 
> > I am still getting the same issue on 2012.10-rc1.
> > 
> > Is anyone else having issues to load a kernel with 2012.10-rc1 on
> > sabrelite?
> 
> Hi Fabio,
> 
> I'm having some trouble running with cache enabled (on Dirk's branch),
> but I wasn't having issues without it
> 
> I haven't pulled 2012.10-rc1 and won't be able to test until
> late in the week (traveling).

What issues ? I've been poking this board when I've been debuging FEC/cache 
issues and that seems clean already.

> Regards,
> 
> 
> Eric

Best regards,
Marek Vasut

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Tom Rini,

> On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
> > On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini <trini@ti.com> wrote:
> > > I lmgtfy.com'd Marek, but for everyone,
> > > http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the
> > > start of the thread.
> > 
> > I am still getting the same issue on 2012.10-rc1.
> > 
> > Is anyone else having issues to load a kernel with 2012.10-rc1 on
> > sabrelite?
> 
> I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
> on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
> Albert, what do you want to do about this?

Whine!

It's broken for me. I've been digging further on and it seems that the crash in 
Fabio's case is caused by:

common/fdt_support.c : fdt_fixup_ethernet() loading char mac[16] = "ethaddr";

where the "ethaddr" string constant is stored unaligned in the .rodata section.

Best regards,
Marek Vasut

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Eric Nelson,

> On 09/24/2012 05:22 PM, Tom Rini wrote:
> > On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
> >> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
> >>> I lmgtfy.com'd Marek, but for everyone,
> >>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the
> >>> start of the thread.
> >> 
> >> I am still getting the same issue on 2012.10-rc1.
> >> 
> >> Is anyone else having issues to load a kernel with 2012.10-rc1 on
> >> sabrelite?
> > 
> > I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
> > on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
> > Albert, what do you want to do about this?
> 
> That adds up, since I've been building with an old one (4.4) and
> haven't been seeing resets.

For starters (as proof-of-concept) ... try this patch ...

diff --git a/common/fdt_support.c b/common/fdt_support.c
index 593f16c..63a5999 100644
--- a/common/fdt_support.c
+++ b/common/fdt_support.c
@@ -449,10 +449,12 @@ void fdt_fixup_ethernet(void *fdt)
 {
        int node, i, j;
        char enet[16], *tmp, *end;
-       char mac[16] = "ethaddr";
+       char mac[16];
        const char *path;
        unsigned char mac_addr[6];
 
+       sprintf(mac, "ethaddr");
+
        node = fdt_path_offset(fdt, "/aliases");
        if (node < 0)
                return;

Best regards,
Marek Vasut

[U-Boot] Resets on mx6qsabrelite

[Fabio Estevam]

On Mon, Sep 24, 2012 at 10:00 PM, Marek Vasut <marex@denx.de> wrote:
> Dear Eric Nelson,
>
>> On 09/24/2012 05:22 PM, Tom Rini wrote:
>> > On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
>> >> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
>> >>> I lmgtfy.com'd Marek, but for everyone,
>> >>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the
>> >>> start of the thread.
>> >>
>> >> I am still getting the same issue on 2012.10-rc1.
>> >>
>> >> Is anyone else having issues to load a kernel with 2012.10-rc1 on
>> >> sabrelite?
>> >
>> > I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
>> > on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
>> > Albert, what do you want to do about this?
>>
>> That adds up, since I've been building with an old one (4.4) and
>> haven't been seeing resets.
>
> For starters (as proof-of-concept) ... try this patch ...

With this patch I can get U-boot 2012.10-rc1 (built with
arm-linux-gnueabi-gcc (Ubuntu/Linaro 4.6.3-1ubuntu5)) to load the
kernel.

Using gcc 4.4 I don't see any problem with 2012.10-rc1.

Regards,

Fabio Estevam

[U-Boot] Resets on mx6qsabrelite

[Sinan Akman]

On 09/24/2012 08:42 PM, Eric Nelson wrote:
> On 09/24/2012 05:22 PM, Tom Rini wrote:
>> On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
>>> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
>>>
>>>> I lmgtfy.com'd Marek, but for everyone,
>>>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the
>>>> start
>>>> of the thread.
>>>
>>> I am still getting the same issue on 2012.10-rc1.
>>>
>>> Is anyone else having issues to load a kernel with 2012.10-rc1 on
>>> sabrelite?
>>
>> I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
>> on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
>> Albert, what do you want to do about this?
>>
>
> That adds up, since I've been building with an old one (4.4) and
> haven't been seeing resets.
>

FWIW Same here, Nitrogen6X, gcc 4.4.6 based toolchain, 2012.10-rc1
no resets.

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Sinan Akman,

> On 09/24/2012 08:42 PM, Eric Nelson wrote:
> > On 09/24/2012 05:22 PM, Tom Rini wrote:
> >> On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
> >>> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>  wrote:
> >>>> I lmgtfy.com'd Marek, but for everyone,
> >>>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html is the
> >>>> start
> >>>> of the thread.
> >>> 
> >>> I am still getting the same issue on 2012.10-rc1.
> >>> 
> >>> Is anyone else having issues to load a kernel with 2012.10-rc1 on
> >>> sabrelite?
> >> 
> >> I bet everyone with a sufficiently recent toolchain is.  I'm seeing this
> >> on TI boards when I use the Linaro toolchain rather than ELDK-5.2.
> >> Albert, what do you want to do about this?
> > 
> > That adds up, since I've been building with an old one (4.4) and
> > haven't been seeing resets.
> 
> FWIW Same here, Nitrogen6X, gcc 4.4.6 based toolchain, 2012.10-rc1
> no resets.

That's because these ancient (and crap ... everything between 4.2 and 4.6 was 
quite crap on arm) toolchains don't generate unaligned string starts in .rodata 
. New toolchains do, yet I have no idea how to fix it ... probably Albert might 
know?

Best regards,
Marek Vasut

[U-Boot] Resets on mx6qsabrelite

[Sinan Akman]

   Hi Marek

On 09/24/2012 11:03 PM, Marek Vasut wrote:
> That's because these ancient (and crap ... everything between 4.2 and 4.6 was
> quite crap on arm) toolchains don't generate unaligned string starts in .rodata
> . New toolchains do, yet I have no idea how to fix it ... probably Albert might
> know?

   I just compiled with ELDK-5.2 toolchain :

U-Boot 2012.10-rc1 (Sep 24 2012 - 23:15:50) 

arm-linux-gnueabi-gcc (GCC) 4.6.4 20120303 (prerelease) 

GNU ld (GNU Binutils) 2.22

I see no resets. So I am not sure if that
means toolchains until including 4.6.4 has
this bug.

Hope this helps.

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Sinan Akman,

>    Hi Marek
> 
> On 09/24/2012 11:03 PM, Marek Vasut wrote:
> > That's because these ancient (and crap ... everything between 4.2 and 4.6
> > was quite crap on arm) toolchains don't generate unaligned string starts
> > in .rodata . New toolchains do, yet I have no idea how to fix it ...
> > probably Albert might know?
> 
>    I just compiled with ELDK-5.2 toolchain :
> 
> U-Boot 2012.10-rc1 (Sep 24 2012 - 23:15:50)
> 
> arm-linux-gnueabi-gcc (GCC) 4.6.4 20120303 (prerelease)
> 
> GNU ld (GNU Binutils) 2.22
> 
> I see no resets. So I am not sure if that
> means toolchains until including 4.6.4 has
> this bug.
> 
> Hope this helps.

It's not a bug, it's feature ... and it was introduced in 4.7* ;-)

[*] I believe Linaro does a lot of backports into their toolchain and other 
random crap, that's why I try to avoid their toolchains as much as possible.

Best regards,
Marek Vasut

[U-Boot] Resets on mx6qsabrelite

[Tom Rini]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 09/24/12 18:00, Marek Vasut wrote:
> Dear Eric Nelson,
> 
>> On 09/24/2012 05:22 PM, Tom Rini wrote:
>>> On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
>>>> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>
>>>> wrote:
>>>>> I lmgtfy.com'd Marek, but for everyone, 
>>>>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html
>>>>> is the start of the thread.
>>>> 
>>>> I am still getting the same issue on 2012.10-rc1.
>>>> 
>>>> Is anyone else having issues to load a kernel with
>>>> 2012.10-rc1 on sabrelite?
>>> 
>>> I bet everyone with a sufficiently recent toolchain is.  I'm
>>> seeing this on TI boards when I use the Linaro toolchain rather
>>> than ELDK-5.2. Albert, what do you want to do about this?
>> 
>> That adds up, since I've been building with an old one (4.4) and 
>> haven't been seeing resets.
> 
> For starters (as proof-of-concept) ... try this patch ...
> 
> diff --git a/common/fdt_support.c b/common/fdt_support.c index
> 593f16c..63a5999 100644 --- a/common/fdt_support.c +++
> b/common/fdt_support.c @@ -449,10 +449,12 @@ void
> fdt_fixup_ethernet(void *fdt) { int node, i, j; char enet[16],
> *tmp, *end; -       char mac[16] = "ethaddr"; +       char
> mac[16]; const char *path; unsigned char mac_addr[6];
> 
> +       sprintf(mac, "ethaddr"); + node = fdt_path_offset(fdt,
> "/aliases"); if (node < 0) return;

This, and a similar patch to fs/ubifs/super.c are the right way to go,
I think.  Please submit this properly, thanks!

- -- 
Tom

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://www.enigmail.net/
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=sNN7
-----END PGP SIGNATURE-----

[U-Boot] Resets on mx6qsabrelite

[Marek Vasut]

Dear Tom Rini,

> On 09/24/12 18:00, Marek Vasut wrote:
> > Dear Eric Nelson,
> > 
> >> On 09/24/2012 05:22 PM, Tom Rini wrote:
> >>> On Mon, Sep 24, 2012 at 09:03:27PM -0300, Fabio Estevam wrote:
> >>>> On Wed, Sep 19, 2012 at 5:20 PM, Tom Rini<trini@ti.com>
> >>>> 
> >>>> wrote:
> >>>>> I lmgtfy.com'd Marek, but for everyone,
> >>>>> http://lists.denx.de/pipermail/u-boot/2012-July/127260.html
> >>>>> is the start of the thread.
> >>>> 
> >>>> I am still getting the same issue on 2012.10-rc1.
> >>>> 
> >>>> Is anyone else having issues to load a kernel with
> >>>> 2012.10-rc1 on sabrelite?
> >>> 
> >>> I bet everyone with a sufficiently recent toolchain is.  I'm
> >>> seeing this on TI boards when I use the Linaro toolchain rather
> >>> than ELDK-5.2. Albert, what do you want to do about this?
> >> 
> >> That adds up, since I've been building with an old one (4.4) and
> >> haven't been seeing resets.
> > 
> > For starters (as proof-of-concept) ... try this patch ...
> > 
> > diff --git a/common/fdt_support.c b/common/fdt_support.c index
> > 593f16c..63a5999 100644 --- a/common/fdt_support.c +++
> > b/common/fdt_support.c @@ -449,10 +449,12 @@ void
> > fdt_fixup_ethernet(void *fdt) { int node, i, j; char enet[16],
> > *tmp, *end; -       char mac[16] = "ethaddr"; +       char
> > mac[16]; const char *path; unsigned char mac_addr[6];
> > 
> > +       sprintf(mac, "ethaddr"); + node = fdt_path_offset(fdt,
> > "/aliases"); if (node < 0) return;
> 
> This, and a similar patch to fs/ubifs/super.c are the right way to go,
> I think.  Please submit this properly, thanks!

I don't think so ... I'm still reading the C spec, give me some time, it's not 
exactly your usual good night light reading, you know ...

Best regards,
Marek Vasut